A method for inverting temperature-dependent mechanical properties of ceramic matrix composites

Abstract

The application discloses a kind of ceramic matrix composite temperature-related mechanical properties inversion method, specifically related to the field of composite material performance characterization.It includes: the ceramic matrix composite to be measured is divided into multiple test pieces;Each test piece is heated to target temperature, and the measured thermal strain of multiple points on the surface of each test piece is obtained;For each test piece, at its target temperature, the test piece is uniaxially stretched by loading end, and the measured tensile strain of multiple points on the surface of the test piece during uniaxial stretching, the measured stress-strain curve is obtained;A parameterized function of the mechanical property parameter vector of the ceramic matrix composite to be measured changes with temperature is constructed, to minimize the error function as the goal, the hyperparameter vector is inverted and identified, to obtain the optimal hyperparameter vector, and the mechanical property parameter vector of the ceramic matrix composite to be measured is obtained in combination with the parameterized function.Based on the above method, the mechanical property parameters of microscale component materials can be obtained.

Description

Technical Field

[0001] This application relates to the field of composite material performance characterization, and in particular to an inversion method for the temperature-dependent mechanical properties of ceramic matrix composites. Background Technology

[0002] Ceramic matrix composites (CMCs) are materials composed of fiber bundles and a matrix, exhibiting complex multi-scale structures and mechanical properties. They hold great promise for applications in extreme environments, such as hot-end components and thermal protection systems in aerospace equipment. However, the complex manufacturing process and inherent multi-scale heterogeneous structure of CMCs result in highly complex anisotropic mechanical properties. The influence of high-temperature environments on the microscopic compositional properties further complicates the temperature-dependent mechanical properties of CMCs, significantly limiting their engineering applications in high-temperature components.

[0003] Traditional measurement techniques are insufficient for efficiently and comprehensively measuring the anisotropic micromechanical properties of ceramic matrix composites. They can only obtain limited macroscopic performance parameters through tensile, compression, and shear experiments, failing to capture the mechanical properties of the microscopic components. Furthermore, the difficulty of high-temperature environmental experiments increases measurement costs, making it impossible to obtain temperature-dependent mechanical properties of ceramic matrix composites. Summary of the Invention

[0004] The main objective of this application is to provide an inversion method for the temperature-dependent mechanical properties of ceramic matrix composites, aiming to solve the problem that existing methods cannot obtain the mechanical property parameters of micro-scale component materials.

[0005] To achieve the above objectives, this application provides a method for inverting the temperature-dependent mechanical properties of ceramic matrix composites, comprising: dividing the ceramic matrix composite to be tested into multiple specimens; heating each specimen to a target temperature, wherein the target temperature for each specimen is different; acquiring measured thermal strain at multiple points on the surface of each specimen at the corresponding target temperature; for each specimen, subjecting it to uniaxial tension at its target temperature through a loading end, and acquiring measured tensile strain at multiple points on the surface of the specimen during uniaxial tension, as well as the load and displacement at the loading end, and determining a measured stress-strain curve based on the load, displacement, and measured tensile strain; and constructing the mechanical properties of the ceramic matrix composite to be tested. A parameterization function is used to transform the hyperparameter vector with temperature, where the parameterization function includes temperature and hyperparameter vectors as coefficients. The hyperparameter vector is inverted and identified with the goal of minimizing the error function to obtain the optimal hyperparameter vector. The optimal hyperparameter vector and temperature are input into the parameterization function to obtain the mechanical property parameter vector of the ceramic matrix composite material under test. The error function is determined based on the differences between predicted and measured thermal strain, predicted and measured stress-strain curves, and predicted tensile strain. The predicted thermal strain, predicted stress-strain curves, and predicted tensile strain are obtained by inputting the hyperparameter vector into a pre-trained prediction model.

[0006] Optionally, the parameterized function is:

[0007]

[0008] In the formula, The first in the mechanical performance parameter vector i One mechanical property parameter, T For temperature, A i , B i , C i All of these are coefficients, which together constitute the hyperparameter vector.

[0009] Optionally, the error function is:

[0010]

[0011] In the formula, For hyperparameter vectors, ω 1. ω 2. ω 3 are all weighting coefficients. This is the thermal expansion error term. This is a macroscopic mechanical error term. This represents the error term of the local strain field.

[0012] Optionally, the thermal expansion error term is determined by the difference between the predicted thermal strain and the measured thermal strain:

[0013]

[0014] In the formula, For the first k The surface of the first test piece j Measured thermal strain at each point T For temperature, For the first k The surface of the first test piece j Predicted thermal strain at each point.

[0015] Optionally, the macroscopic mechanical error term is determined by the difference between the predicted stress-strain curve and the measured stress-strain curve:

[0016]

[0017] In the formula, For the first k The measured stress-strain curves of each specimen. For the first k Predicted stress-strain curves for each specimen For the first k The maximum stress in the measured stress-strain curve of each specimen; For each strain ε The integral of the difference between the measured stress-strain curve and the predicted stress-strain curve at the point.

[0018] Optionally, the local strain field error term is determined by the difference between the predicted tensile strain and the measured tensile strain:

[0019]

[0020] In the formula, For the first k The surface of the first test piece j Measured tensile strain at each point For the first k The surface of the first test piece j Predicted tensile strain at each point.

[0021] Optionally, with the goal of minimizing the error function, the hyperparameter vector is inverted and identified to obtain the optimal hyperparameter vector, including: treating the hyperparameter vector as an individual, calculating the fitness value according to the error function, iteratively optimizing the hyperparameter vector through the differential evolution algorithm to obtain the optimization result; and using the Nelder-Mead algorithm to optimize the optimization result to obtain the optimal hyperparameter vector.

[0022] Optionally, the training process of the pre-trained prediction model includes: establishing a finite element model of the ceramic matrix composite material to be tested, and assigning material properties to the fiber bundles and matrix of the finite element model respectively; sampling hyperparameter vectors within a preset range to obtain multiple hyperparameter vectors; inputting each hyperparameter vector into the finite element model of the ceramic matrix composite material to be tested for finite element calculation to obtain thermal strain, stress-strain curves and tensile strain; using the hyperparameter vectors as input and the corresponding thermal strain, stress-strain curves and tensile strain as outputs to train the neural network surrogate model to obtain the pre-trained prediction model.

[0023] To achieve the above objectives, this application also provides an inversion device for the temperature-dependent mechanical properties of ceramic matrix composites, including an experimental measurement module for dividing the ceramic matrix composite to be tested into multiple specimens; heating each specimen to a target temperature, wherein the target temperature for each specimen is different; acquiring the measured thermal strain at multiple points on the surface of each specimen at the corresponding target temperature; for each specimen, uniaxially stretching it through a loading end at its target temperature, and acquiring the measured tensile strain at multiple points on the surface of the specimen during the uniaxial stretching process, as well as the load and displacement at the loading end, and determining the measured stress-strain curve based on the load, displacement, and measured tensile strain; and a parameterization function construction module for constructing the measured ceramic matrix composite... The test involves a parameterization function that represents the change of the mechanical property parameter vector of the composite material with temperature. This parameterization function includes temperature and a hyperparameter vector as coefficients. An inversion identification module, aiming to minimize the error function, performs inversion identification on the hyperparameter vector to obtain the optimal hyperparameter vector. The optimal hyperparameter vector and temperature are then input into the parameterization function to obtain the mechanical property parameter vector of the ceramic matrix composite material under test. The error function is determined based on the differences between predicted and measured thermal strain, predicted and measured stress-strain curves, and predicted tensile strain. The predicted thermal strain, predicted stress-strain curves, and predicted tensile strain are obtained by inputting the hyperparameter vector into a pre-trained prediction model.

[0024] Compared with the prior art, the beneficial effects of this application are as follows:

[0025] The method for inverting the temperature-related mechanical properties of ceramic matrix composites of the present invention involves dividing the ceramic matrix composite to be tested into multiple specimens, obtaining the measured thermal strain of each specimen at the corresponding target temperature, and obtaining the measured tensile strain and measured stress-strain curves of the specimen surface during uniaxial tensile testing.

[0026] A parameterized function is constructed to represent the temperature-dependent changes in the mechanical property parameters of the ceramic matrix composite material under test. Based on the differences between predicted and measured thermal strain, predicted and measured stress-strain curves, and predicted and measured tensile strain, the hyperparameter vectors in the parameterized function are inverted and identified, thus obtaining the mechanical property parameters of the micro-scale components. In this process, the hyperparameters obtained by combining the errors of thermal strain, stress-strain curves, and tensile strain are taken into account. The thermal strain and tensile strain are measured only through the measurement system, eliminating the need to directly measure the mechanical property parameters under high-temperature conditions, thereby reducing measurement costs. Attached Figure Description

[0027] Figure 1 This is a flowchart illustrating the method for inverting the temperature-dependent mechanical properties of ceramic matrix composites according to this application.

[0028] Figure 2 This is a measured thermal strain field diagram of the specimen surface at 800°C obtained in Example 1 of the method for inverting the temperature-related mechanical properties of ceramic matrix composites according to this application.

[0029] Figure 3 This is an example of an inversion method for temperature-related mechanical properties of ceramic matrix composites according to this application, showing the measured tensile strain field on the surface of a specimen at 800°C.

[0030] Figure 4 This is a measured stress-strain curve at 800°C obtained from Example 1 of the method for inverting the temperature-related mechanical properties of ceramic matrix composites according to this application.

[0031] Figure 5 This is a comparison diagram between the stress-strain curve obtained by the inversion method for the temperature-related mechanical properties of ceramic matrix composites in this application and the empirical stress-strain curve.

[0032] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0034] The first embodiment of the present invention provides a method for inverting the temperature-dependent mechanical properties of ceramic matrix composites, such as... Figure 1As shown, the specific steps include:

[0035] Step S1: The ceramic matrix composite material to be tested is divided into multiple specimens; each specimen is heated to a target temperature, wherein the target temperature for each specimen is different.

[0036] Step S2: At the corresponding target temperature, obtain the measured thermal strain at multiple points on the surface of each specimen caused by thermal expansion.

[0037] Step S3: For each specimen, at its target temperature, uniaxial tension is applied to the specimen through the loading end, and the measured tensile strain at multiple points on the surface of the specimen, as well as the load and displacement at the loading end, are obtained during the uniaxial tension process. The measured stress-strain curve is determined based on the load, displacement, and measured tensile strain.

[0038] Step S4: Construct a parameterized function for the variation of the mechanical property parameter vector of the ceramic matrix composite material under test with temperature. The parameterized function includes temperature and a hyperparameter vector as coefficients. The parameterized function is as follows:

[0039]

[0040] In the formula, The first in the mechanical performance parameter vector i One mechanical property parameter, T For temperature, A i , B i , C i All of these are coefficients, which together constitute the hyperparameter vector.

[0041] It is worth noting that the parameterization function is set based on knowledge and experience in the field of ceramic matrix composites. It stipulates that the target mechanical property parameters change with temperature in a quadratic function form. This is mainly because the mechanical property parameters of ceramic matrix composite components generally exhibit a monotonic nonlinear increase or decrease with increasing temperature, and there are no very complex changes. Therefore, the quadratic function form is sufficient to describe the parameter change law and can meet the accuracy requirements of engineering calculations.

[0042] In this embodiment, the mechanical performance parameter vector includes at least the matrix elastic modulus. Matrix thermal expansion coefficient Longitudinal elastic modulus of fiber bundle transverse elastic modulus of fiber bundle and the longitudinal thermal expansion coefficient of the fiber bundle ,Right now In this model, subscript 1 represents the x-axis direction in space, subscript 2 represents the y-axis direction, and subscript 3 represents the z-axis direction. In this embodiment, the x-axis direction is the fiber bundle axial direction, and the y-axis direction is the fiber bundle radial direction. The main reason for selecting these five mechanical performance parameters is that the elastic modulus of the matrix, the transverse elastic modulus of the fiber bundle, and the longitudinal elastic modulus of the fiber bundle change relatively significantly with temperature, and are the main factors affecting the mechanical properties of ceramic matrix composites at high temperatures. The other elastic parameters (Poisson's ratio of the fiber bundle and the matrix, shear modulus, etc.) change very weakly with temperature and can be considered to be fixed with temperature. In addition to the elastic parameters, the thermal expansion coefficient of the matrix and the thermal expansion coefficient of the fiber bundle are another major factor affecting the high-temperature performance of ceramic matrix composites. The thermal expansion coefficient of the fiber bundle can be considered to be approximately isotropic, so only the longitudinal thermal expansion coefficient of the fiber bundle is selected as the target. Therefore, the hyperparameter vector is... θ =( A 1, A 2,…… A 5, B 1, B 2,…… B 5, C 1, C 2,……, C 5).

[0043] Step S5: With the goal of minimizing the error function, the hyperparameter vector is inverted and identified to obtain the optimal hyperparameter vector;

[0044] Specifically, the hyperparameter vector is treated as an individual, and its fitness value is calculated based on the error function. A differential evolution algorithm is used to globally optimize the hyperparameter vector, yielding preliminary optimization results. The Nelder-Mead algorithm is then used to locally optimize these results, resulting in the optimal hyperparameter vector. Further, the hyperparameter vector is treated as an individual, and its value range is set to randomly generate an initial population. Mutation and crossover operations are performed on the population to obtain individuals for comparison. These individuals are input into a pre-trained prediction model to obtain predicted thermal strain, predicted stress-strain curves, and predicted tensile strain. Fitness values ​​are calculated, and a new generation of population is obtained by minimizing the fitness value. Iterative optimization is performed through these steps until convergence is achieved, yielding preliminary optimization results. These results are used as the initial values ​​for the Nelder-Mead algorithm, which is then used to further optimize the hyperparameter vector until the algorithm reaches convergence, resulting in the optimal hyperparameter vector. A 1, A 2,…… A 5, B 1, B 2,…… B 5, C 1, C2,……, C 5).

[0045] The error function is determined based on the differences between the predicted and measured thermal strain, the differences between the predicted and measured stress-strain curves, and the differences between the predicted and measured tensile strain. The error function is:

[0046]

[0047] In the formula, For hyperparameter vectors, ω 1. ω 2. ω 3 are all weighting coefficients. This is the thermal expansion error term. This is a macroscopic mechanical error term. This represents the error term of the local strain field.

[0048] The thermal expansion error term is determined by the difference between the predicted thermal strain and the measured thermal strain:

[0049]

[0050] In the formula, For the first k The first surface of the test piece j Measured thermal strain at each point T For temperature, For the first k The first surface of the test piece j Predicted thermal strain at each point.

[0051] The macroscopic mechanical error term is determined by the difference between the predicted stress-strain curve and the measured stress-strain curve:

[0052]

[0053] In the formula, For the first k The measured stress-strain curves of each specimen. For the first k The predicted stress-strain curves of each specimen. For the first k The maximum stress in the measured stress-strain curve of each specimen is located in the denominator and used for normalization. For each strain ε The integral of the difference between the measured stress-strain curve and the predicted stress-strain curve at the point.

[0054] The local strain field error term is determined by the difference between the predicted tensile strain and the measured tensile strain:

[0055]

[0056] In the formula, For the first k The surface of the first test piece j Measured tensile strain at each point For the first k The surface of the first test piece j Predicted tensile strain at each point.

[0057] Predicting thermal strain, stress-strain curves, and tensile strain is achieved by inputting hyperparameter vectors into a pre-trained prediction model. The training process of the pre-trained prediction model is as follows:

[0058] A finite element model of the ceramic matrix composite material to be tested is established, and the fiber bundles and matrix of the finite element model are assigned material properties respectively. Hyperparameter vectors are sampled within a preset range to obtain multiple hyperparameter vectors. Mechanical property parameter vectors are calculated according to the parameterization function, ensuring they fall within the range of values ​​for each mechanical property parameter. Each mechanical property parameter vector is input into the finite element model of the ceramic matrix composite material to be tested for finite element calculation to obtain thermal strain, stress-strain curves, and tensile strain. The mechanical property parameter vectors are used as input, and the corresponding thermal strain, stress-strain curves, and tensile strain are used as outputs to train a neural network surrogate model, resulting in a pre-trained prediction model.

[0059] It's worth noting that for neural network surrogate models predicting thermal strain, stress-strain curves, and tensile strain, a multi-task ResUNet network can be chosen. This network uses a vector of mechanical property parameters as input and simultaneously outputs thermal strain, stress-strain curves, and tensile strain for training, resulting in a predictive model that can simultaneously predict thermal strain, stress-strain curves, and tensile strain. Alternatively, three neural network surrogate models can be selected. For example, a fully connected neural network (MLP) can be built for predicting stress-strain curves. The input layer of this network is a vector of mechanical property parameters, and the output layer contains 30 nodes, corresponding to 30 equally spaced data points on the predicted stress-strain curve. For predicting thermal strain, a convolutional neural network (CNN) can be built. The input layer of this network is a vector of mechanical property parameters, and the output layer is a two-dimensional thermal strain field image (resolution: [resolution value missing]). The corresponding measurement area in the middle section of the specimen (Strain data from DIC measurement points). To predict tensile strain, a convolutional neural network (CNN) is built. The input layer of this network is a vector of mechanical property parameters, and the output layer is a two-dimensional image of the tensile strain field.

[0060] Step S6: Combine the optimal hyperparameter vector with the temperature input parameterization function to obtain the mechanical property parameter vector of the ceramic matrix composite material under test.

[0061] A second embodiment of the present invention provides an inversion device for the temperature-dependent mechanical properties of ceramic matrix composites, comprising: an experimental measurement module for dividing the ceramic matrix composite to be tested into multiple specimens; heating each specimen to a target temperature, wherein the target temperature for each specimen is different; acquiring measured thermal strain at multiple points on the surface of each specimen at the corresponding target temperature; for each specimen, uniaxially stretching the specimen through a loading end at its target temperature, and acquiring measured tensile strain at multiple points on the surface of the specimen during uniaxial stretching, as well as the load and displacement at the loading end, and determining a measured stress-strain curve based on the load, displacement, and measured tensile strain;

[0062] In this embodiment, the experimental measurement module includes a testing machine and a non-contact full-field strain measurement DIC (Digital Image Correlation) system. The non-contact full-field strain measurement DIC system can realize the measured thermal strain at multiple points on the surface of each specimen and the measured tensile strain at multiple points on the surface of the specimen during uniaxial tensile testing. The specimen is subjected to uniaxial tensile testing through the loading end of the testing machine, and the load-displacement curve is recorded.

[0063] The parameterization function construction module is used to construct a parameterization function that shows the change of the mechanical property parameter vector of the ceramic matrix composite material under test with temperature. The parameterization function includes temperature and hyperparameter vectors as coefficients.

[0064] The inversion identification module aims to minimize the error function by inverting and identifying the hyperparameter vector to obtain the optimal hyperparameter vector. The optimal hyperparameter vector is then combined with the temperature input parameterization function to obtain the mechanical property parameter vector of the ceramic matrix composite material under test.

[0065] The error function is determined based on the difference between the predicted thermal strain and the measured thermal strain, the difference between the predicted stress-strain curve and the measured stress-strain curve, and the difference between the predicted tensile strain and the measured tensile strain.

[0066] Predicting thermal strain, stress-strain curves, and tensile strain is achieved by inputting hyperparameter vectors into a pre-trained prediction model.

[0067] Example 1

[0068] Two-dimensional braided SiC f / SiC m Ceramic matrix composites were used as the ceramic matrix composites to be tested.

[0069] Step S10: The ceramic matrix composite material to be tested is segmented to obtain four SiC atoms. f / SiC mSpecimens; four specimens were heated to target temperatures, including 400°C, 800°C, 1000°C, and 1200°C, respectively. k =1, 2, 3, 4;

[0070] Step S20: At the corresponding target temperature, obtain the measured thermal strain at multiple points on the surface of each specimen due to thermal expansion; wherein, the measured thermal strain on the specimen surface at 800°C is shown in [reference needed]. Figure 2 ;

[0071] Step S30: For each specimen, at its target temperature, the tensile loading rate is set to 0.5 mm / min. Uniaxial tension is applied to the specimen through the loading end, and the measured tensile strain at multiple points on the specimen surface is obtained when the displacement loading is 0.2 mm. The load and displacement at the loading end are also recorded. The measured stress-strain curve is determined based on the load, displacement, and measured tensile strain. The measured tensile strain on the specimen surface at 800°C is shown in [reference needed]. Figure 3 At this time, SiC f / SiC m The specimen remained within its elastic response range, and no significant damage occurred within the material. The material was then loaded to fracture at this tensile rate, and the measured stress-strain curve was obtained as follows: Figure 4 As shown.

[0072] Step S40: Using the differential evolution algorithm, the population generation is set to 120, and the number of individuals in each generation is 150. A 1, A 2,…… A 5, B 1, B 2,…… B 5, C 1, C 2,……, C 5) For each individual, calculate the fitness value based on the error function, output the optimized individual, and use it as the initial value for the Nelder-Mead algorithm for further optimization until the algorithm reaches the convergence criterion and outputs the optimal hyperparameter vector. During the optimization process, the reasonable value ranges for the five mechanical performance parameters are shown in Table 1. The above two algorithms are used to optimize the hyperparameter vector ( A 1, A 2,…… A 5, B 1, B 2,…… B 5, C 1, C 2,……, C5) During iterative optimization, the hyperparameter vector is checked at each iteration step through the parameterization function. That is, the hyperparameter vector is substituted into the parameterization function to calculate the mechanical performance parameters. If the combination of its values ​​causes the mechanical performance parameters to exceed the reasonable range, the iterative value of the hyperparameter vector will be updated to ensure that the mechanical performance parameters do not exceed the range.

[0073] Table 1 Range of Mechanical Performance Parameters

[0074]

[0075] Step S6: Combine the optimal hyperparameter vector with the temperature input parameterization function to obtain the mechanical property parameter vector of the ceramic matrix composite material under test, as shown in Table 2:

[0076] Table 2 Mechanical property parameters of the ceramic matrix composites to be tested

[0077]

[0078] Organizing SiC f / SiC m Classic literature and materials in the field of performance research yielded a set of SiC f / SiC m The empirical rules for the five mechanical performance parameters are as follows:

[0079] Table 3 SiC f / SiC m Empirical formulas for the variation of mechanical property parameters with temperature

[0080]

[0081] Substituting the temperature variation patterns of the parameters in Tables 2 and 3 into the finite element numerical model, we obtain the SiC... f / SiC m Comparison of the inverse stress-strain curves (Table 2), empirical stress-strain curves (Table 3), and measured stress-strain curves (800°C) of the material. Figure 5 As shown, the stress-strain curve derived from this embodiment (calculated curve based on inversion values) fits the measured stress-strain curve (800°C experimental data) better, while the empirical stress-strain curve (calculated curve based on empirical values) deviates more from the measured stress-strain curve. This is mainly because empirical curves for these mechanical property parameters often lack data for SiC. f / SiC m The consideration of factors such as porosity defects caused by the actual material preparation process results in the elastic parameters such as the modulus of the material at room temperature and high temperature being higher than the actual values.

[0082] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for inverting the temperature-dependent mechanical properties of ceramic matrix composites, characterized in that, include: The ceramic matrix composite material to be tested was segmented to obtain multiple specimens; Each specimen is heated to a target temperature, where the target temperature is different for each specimen. At the corresponding target temperature, the measured thermal strain at multiple points on the surface of each specimen was obtained; For each specimen, at its target temperature, the specimen is subjected to uniaxial tension through the loading end, and the measured tensile strain at multiple points on the surface of the specimen, as well as the load and displacement at the loading end, are obtained during the uniaxial tension process. The measured stress-strain curve is determined based on the load, displacement, and measured tensile strain. A parameterized function is constructed to represent the change of the mechanical property parameter vector of the ceramic matrix composite material under test with temperature. The parameterized function includes temperature and hyperparameter vector as coefficients. The mechanical property parameter vector includes the matrix elastic modulus, the matrix thermal expansion coefficient, the longitudinal elastic modulus of the fiber bundle, the transverse elastic modulus of the fiber bundle, and the longitudinal thermal expansion coefficient of the fiber bundle. With the goal of minimizing the error function, the hyperparameter vector is inverted and identified to obtain the optimal hyperparameter vector; By combining the optimal hyperparameter vector with the temperature input parameterization function, the mechanical property parameter vector of the ceramic matrix composite material under test can be obtained. The error function is determined based on the difference between the predicted thermal strain and the measured thermal strain, the difference between the predicted stress-strain curve and the measured stress-strain curve, and the difference between the predicted tensile strain and the measured tensile strain. Predicting thermal strain, stress-strain curves, and tensile strain is achieved by inputting hyperparameter vectors into a pre-trained prediction model. The parameterized function is: In the formula, The first in the mechanical performance parameter vector i One mechanical property parameter, where T is temperature. A i , B i , C i All of these are coefficients, which together constitute the hyperparameter vector.

2. The method for inverting the temperature-dependent mechanical properties of ceramic matrix composites according to claim 1, characterized in that, The error function is: In the formula, For hyperparameter vectors, ω 1. ω 2. ω 3 are all weighting coefficients. This is the thermal expansion error term. This is a macroscopic mechanical error term. This represents the error term of the local strain field.

3. The method for inverting the temperature-dependent mechanical properties of ceramic matrix composites according to claim 2, characterized in that, The thermal expansion error term is determined by the difference between the predicted thermal strain and the measured thermal strain: In the formula, For the first k The first surface of the test piece j Measured thermal strain at each point T For temperature, For the first k The first surface of the test piece j Predicted thermal strain at each point.

4. The method for inverting the temperature-dependent mechanical properties of ceramic matrix composites according to claim 2, characterized in that, The macroscopic mechanical error term is determined by the difference between the predicted stress-strain curve and the measured stress-strain curve: In the formula, For the first k The measured stress-strain curves of each specimen. For the first k Predicted stress-strain curves for each specimen For the first k The maximum stress in the measured stress-strain curve of each specimen; For each strain ε The integral of the difference between the measured stress-strain curve and the predicted stress-strain curve at the point.

5. The method for inverting the temperature-dependent mechanical properties of ceramic matrix composites according to claim 2, characterized in that, The local strain field error term is determined by the difference between the predicted tensile strain and the measured tensile strain: In the formula, For the first k The first surface of the test piece j Measured tensile strain at each point For the first k The first surface of the test piece j Predicted tensile strain at each point.

6. The method for inverting the temperature-dependent mechanical properties of ceramic matrix composites according to claim 1, characterized in that, With the goal of minimizing the error function, the hyperparameter vector is inverted and identified to obtain the optimal hyperparameter vector, including: Treating the hyperparameter vector as an individual, the fitness value is calculated based on the error function. The hyperparameter vector is then iteratively optimized using the differential evolution algorithm to obtain the optimization result. Finally, the Nelder-Mead algorithm is used to further optimize the optimization result to obtain the optimal hyperparameter vector.

7. The method for inverting the temperature-dependent mechanical properties of ceramic matrix composites according to claim 1, characterized in that, The training process of a pre-trained prediction model includes: A finite element model of the ceramic matrix composite material to be tested is established, and the material properties of the fiber bundles and the matrix are assigned to the finite element model respectively. Sample the hyperparameter vectors within a preset range to obtain multiple hyperparameter vectors; Each hyperparameter vector is input into the finite element model of the ceramic matrix composite material under test for finite element calculation to obtain thermal strain, stress-strain curves and tensile strain; Using the hyperparameter vector as input and the corresponding thermal strain, stress-strain curves, and tensile strain as output, the neural network surrogate model is trained to obtain a pre-trained prediction model.

8. An inversion device for the temperature-dependent mechanical properties of ceramic matrix composites, characterized in that, include: The experimental measurement module is used to divide the ceramic matrix composite material under test into multiple specimens. Each specimen is heated to a target temperature, which is different for each specimen. At the target temperature, the measured thermal strain at multiple points on the surface of each specimen is obtained. For each specimen, at its target temperature, uniaxial tension is applied through the loading end, and the measured tensile strain at multiple points on the surface of the specimen, as well as the load and displacement at the loading end, are obtained during the uniaxial tension process. The measured stress-strain curve is determined based on the load, displacement, and measured tensile strain. The parameterization function construction module is used to construct a parameterization function that shows the change of the mechanical property parameter vector of the ceramic matrix composite material under test with temperature. The parameterization function includes temperature and hyperparameter vectors as coefficients. The inversion identification module aims to minimize the error function by inverting and identifying the hyperparameter vector to obtain the optimal hyperparameter vector. The optimal hyperparameter vector is then combined with the temperature input parameterization function to obtain the mechanical property parameter vector of the ceramic matrix composite material under test. The error function is determined based on the difference between the predicted thermal strain and the measured thermal strain, the difference between the predicted stress-strain curve and the measured stress-strain curve, and the difference between the predicted tensile strain and the measured tensile strain. Predicting thermal strain, stress-strain curves, and tensile strain is achieved by inputting hyperparameter vectors into a pre-trained prediction model.

Images